Proc. Natl. Acad. Sci. USAVol. 92, pp. 6728-6732, July 1995Genetics

Generation of a high-titer retroviral vector capable of expressinghigh levels of the human (8-globin gene

(hemoglobinopathies/gene therapy/locus control region)

MICHEL SADELAIN*t, C. H. JASON WANG*, MICHAEL ANTONIOUt, FRANK GROSVELDt§,AND RICHARD C. MULLIGAN*¶*Whitehead Institute for Biomedical Research and Department of Biology, Massachusetts Institute of Technology, Nine Cambridge Center, Cambridge, MA02142-1479; and $Laboratory for Gene Structure and Expression, Medical Research Council National Institute for Medical Research, The Ridgeway,Mill Hill, London, NW7, United Kingdom

Communicated by David E. Housman, Massachusetts Institute of Technology, Cambridge, MA, December 12, 1994

ABSTRACT Retrovirus-mediated gene transfer into he-matopoietic cells may provide a means of treating bothinherited and acquired diseases involving hematopoietic cells.Implementation of this approach for disorders resulting frommutations affecting the 8-globin gene (e.g., P-thalassemia andsickle cell anemia), however, has been hampered by theinability to generate recombinant viruses able to efficientlyand faithfully transmit the necessary sequences for appropri-ate gene expression. We have addressed this problem bycarefully examining the interactions between retroviral andj3-globin gene sequences which affect vector transmission,stability, and expression. First, we examined the transmissionproperties ofa large number ofdifferent recombinant proviralgenomes which vary both in the precise nature of vector,18-globin structural gene, and locus control region (LCR) coresequences incorporated and in the placement and orientationof those sequences. Through this analysis, we identified onespecific vector, termed Mj36L, which carries both the human,8-globin gene and core elements HS2, HS3, and HS4 from theLCR and faithfully transmits recombinant proviral sequencesto cells with titers greater than 106 per ml. Populations ofmurine erythroleukemia (MEL) cells transduced by this virusexpressed levels ofhuman j8-globin transcript which, on a pergene copy basis, were 78% of the levels detected in anMEL-derived cell line, Hull, which carries human chromo-some 11, the site of the 13-globin locus. Analysis of individualtransduced MEL cell clones, however, indicated that, whileexpression was detected in every clone tested (n = 17), thelevels of human j8-globin treatment varied between 4% and146% of the levels in Hull. This clonal variation in expressionlevels suggests that small ,-globin LCR sequences may notprovide for as strict chromosomal position-independent ex-pression of 18-globin as previously suspected, at least in thecontext of retrovirus-mediated gene transfer.

The successful treatment of f3-globin disorders by gene therapywill likely require the efficient and stable introduction of afunctional ,3-globin gene into self-renewing hematopoieticstem cells. While we and others had established that retroviralvectors encoding the human f-globin structural gene coulddirect the erythroid-specific synthesis of 3-globin in miceengrafted with genetically modified bone marrow cells (1-4),these studies suffered from two important limitations. First,the retroviral titers generally obtained with those vectors madeit difficult to generate long-term bone marrow chimerasengrafted with efficiently transduced cells. A second problemwas the variable and generally low level of human 13-globinexpression in vivo observed in the reconstituted animals.

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

The prospects for increasing the level of f-globin expressionobtainable in retroviral vectors were dramatically improved bythe recent finding that DNA sequences located considerabledistances from the human l3-globin structural gene on chro-mosome 11, termed locus control region (LCR) sequences (5,6), play a critical role in the transcriptional control of geneswithin the P-globin-like gene cluster (reviewed in refs. 7-9).Specifically, transgenic experiments from a number of labo-ratories have indicated that the linkage of LCR sequences tothe human 13-globin structural gene dramatically increases thelevel of expression of 3-globin observed in erythroid cells(10-13). Some studies further suggested that LCR sequencesmay differ from classical transcriptional enhancer sequences inthat they provide for the chromosomal position-independentexpression of linked genes (10-16). Smaller fragments, moresuitable for insertion into vectors, have recently been defined,which have been reported to retain partial LCR activity(17-29). Recent efforts by several groups to incorporate LCRsubfragments into ,3-globin retroviral vectors, however, haveresulted in vector rearrangements (ref. 30 and our observa-tions), poor titers (31), or very low expression of the ,B-globingene (32).We report here a systematic study of the ,B-globin gene, LCR

core sites, and retroviral sequences that control vector trans-mission. This analysis has led to the generation of a high-titer,genomically stable retroviral vector bearing the human ,3-glo-bin gene and the LCR core sites HS2, HS3, and HS4. Thisvector confers elevated and erythroid-specific expression ofthe ,B-globin gene. Expression in different clones of transducedmurine erythroleukemia (MEL) cells is variable, however,raising questions about the ability of small LCR sequences toconfer position-independent gene expression, at least in thecontext of a retroviral vector.

MATERIALS AND METHODSGeneration of Vectors and Packaging Cell Lines. The

retroviral vectors pSG and pMFG are described elsewhere (33,34). The 13-globin gene was subcloned from pSVX-Neo (1).The f3-globin coding sequence (13CS) consists of a 444-bp NcoI-linker-containing ,-globin sequence from the translationalstart to the stop codon derived from the cDNA. The HS2, HS3,and HS4 fragments (see text and refs. 19-22 and 27-29) were,respectively, given HindIII, EcoRI, and Mlu I linkers andsubcloned in all possible orientations in a Sac II-HindIlI-

Abbreviations: LCR, locus control region; LTR, long terminal repeat;MEL, murine erythroleukemia; MLV, murine leukemia virus.tPresent address: Department of Human Genetics, Memorial Sloan-Kettering Cancer Center, Box 182,1275 York Avenue, New York, NY10021.

§Present address: Department of Cell Biology, Erasmus University,Postbus 1738, 3000 Dr Rotterdam, The Netherlands.1To whom reprint requests should be addressed.

6728

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 6729

EcoRI-Mlu I-Sac II polylinker. The vectors described herein,lacking any selectable marker, were introduced by calciumphosphate cotransfection with pSV2Neo (35) into the qi-CREpackaging cell line (36). After 10 days of selection in G418(Geneticin, Sigma) at 1 mg/ml, the G418-resistant colonies(>50 per transfection) were pooled by trypsinization andgrown in Dulbecco's modified Eagle's medium supplementedwith 10% calf serum, penicillin, and streptomycin withoutG418.

Fibroblast Infection and Viral Transmission Assay. Fivemilliliters of supernatant harvested from pooled producer cellswas applied onto 106 NIH 3T3 fibroblasts. Polybrene (Sigma)was added at 8 ,ug/ml. After 4 hr, the conditioned medium was

removed. Three days later, genomic DNA was extracted fromthe confluent NIH 3T3 fibroblasts. For Southern analysis,10-,ug samples of genomic DNA from the producer and thecorresponding 3T3 target cells were digested overnight withthe restriction enzyme Nhe I [which cuts in both long terminalrepeats (LTRs)] and run side by side on a 1% agarose gel.[32P]dCTP-labeled 13-globin probes were generated by usingthe human Nco I-EcoRI genomic template. Band intensity wasquantified by using a phosphorimager (Fuji Bio-Imaging).Vector signal intensity was normalized to that of the endog-enous band. Very even loading in Figs. 1-3 allows for directreading of the gels as shown.MEL Cell Infection and Screening. C88 MEL cells were

grown in RPMI medium/10% fetal bovine serum/penicillin/streptomycin. Infection was performed by a 48-hr cocultivationof 105 MEL cells, treated for the previous 18 hr with tunica-mycin (Sigma) at 0.2 ,ug/ml, on the ecotropic producer in thepresence of Polybrene at 4 ,ug/ml. After cocultivation, MELcells were subcloned onto 96-well plates at 0.2 cell per well.Single clones were then expanded and genomic DNA was

prepared. Infected clones were screened by PCR using theprimers GCAAGAAAGTGCTCGGTG in exon 2 and TCT-GATAGGCAGCCTGCA in exon 3. The DNA of positiveclones was further analyzed by Southern blotting for copy

number and integration site.MEL Cell Induction and Globin mRNA Quantification.

After 4 days in 2% (vol/vol) dimethyl sulfoxide, RNA was

extracted from each MEL cell clone by using the lithiumchloride/urea procedure (37). This induction-extraction pro-

cedure was repeated at least once. RNA was quantified byRNase protection (38) using probes described in ref. 39 forhuman 13-globin and a 180-nucleotide Pst I-BamHI genomicfragment (40) for mouse a-globin. One to 2 ,g of RNA per

extract was incubated with both admixed probes, which, afterRNase digestion, were electrophoresed on a 6% polyacryl-amide/urea gel. Signal intensity for each band (the a-globinserving as internal control for both induction and loading onto

the gel) was determined by using the phosphorimager.

RESULTSFeatures of Retroviral Vector Design Which Affect Trans-

mission of the Human 18-Globin Structural Gene. To assess thetiter and genomic stability of different retroviral vectors

bearing the human ,3-globin gene, we first developed a directassay to characterize and quantitate vector transmission. Thisassay is based on a quantitative Southern blot analysis of vectorcopy number in both transfected packaging cells and infected3T3 fibroblasts. Supernatant is harvested from a pool of stablevirus-producing cell clones and used to infect the target

fibroblasts under defined conditions (see Materials and Meth-ods). Southern blot analysis of vector DNA copy number in thepolyclonal producer and target cells allows for a direct eval-uation of average transmission for any retroviral construct.

This is achieved by direct comparison of signal intensity in bothgenomic DNA extracts run side by side and therefore does notrely on indirect readouts dependent on gene expression (such

as G418 resistance). This method provides not only for a directmeasurement of gene transfer but also for the detection ofrearrangements of any given construction. As shown in Fig. 1,transmission of a minimal 1.8-kb 13-globin gene and promoter(-129 to + 1650) cloned in reverse orientation in the pSGvector (33) was not detectable in this assay (lanes A). Trans-mission of the larger 2.8-kb Sph I-Pst I (-615 to +2163) globinfragment was also not detectable (data not shown). In contrast,the f3-globin coding sequence alone transmitted very well,showing comparable signals in the packaging and target cellDNAs (lanes B). The high titer of the latter vector, however,was greatly reduced by inclusion of the core sites HS2, HS3,and HS4 of the LCR (see below), which were introduced in thevector upstream of the globin sequence in either orientation(lanes C and D).To determine which P-globin genomic sequences were re-

sponsible for the dramatic reduction in titer, we constructed aset of recombinant vectors bearing systematic deletions ofuntranslated regions of the gene. As shown in Fig. 2, deletionof the 5' untranslated region and intron 1, either alone or incombination, failed to increase viral titers (lanes A-D). De-letion of intron 2 (lanes E) increased transmission to levelssimilar to the coding sequence vector (lanes I), yet furthercombined deletions did not increase titer any more (lanesF-H). Because ,B-globin DNA templates lacking intron 2 directpoor ,B-globin expression (refs. 24 and 41-44 and data notshown), we examined transmission of two subintronic dele-tions (from +518 to + 1288 and +679 to + 953). Both yieldedsimilar titers, which were still 10-fold down from the intronlessgene (shown for the first deletion, lanes J, Fig. 2). Therefore,intron 2 alone accounts for reduced titers of 13-globin gene, anddeletions within intron 2, while increasing titer, still transmitless well than intronless constructions.Recombinant Human 3-Globin Genes Yield Higher Titers

in the pMFG Vector. Since decreased viral transmission mayresult from a negative interaction between 13-globin and ret-roviral sequences, which is only partially alleviated by intron 2subdeletions, we compared transmission of /3-globin sequencesin different vectors. Modifications of the length of gag se-quence, replacement of the Moloney murine leukemia virus(MLV) packaging signal by that of Ha-Ras MLV, and muta-tion of the splice donor site present in pSG failed to increasevector transmission (data not shown). Subcloning recombinant,3-globin fragments into the pMFG vector (34), however,significantly increased transmission (Fig. 3). Lanes A and Dshow transmission of the full P-globin fragment; lanes B andE, the genomic sequence bearing the intron 2 deletion shown

0.51.0 A B C D

FIG. 1. Comparison of vector transmission by Southern blot anal-ysis of vector copy number in the packaging cell (left lane) and targetcell (right lane) genomic DNA. In lanes A, 1.8-kb 13-globin gene andpromoter; in lanes B, f3CS; in lanes C and D, f3CS and the three coreelements HS2, HS3, and HS4 in sense and antisense orientation,respectively. See text for sequence description. Genomic DNA wasdigested with Nhe I and blots were probed with a 3-globin probe.Control lanes show signal intensity for one proviral copy per cell (1.0)and one copy per two cells (0.5). Even loading (data not shown) allowsfor directly comparing signal intensity between lanes.

Genetics: Sadelain et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995)

0510 A B C D E F G Hri ii I i r r- 1!

.S5m . . ..5S-inait_ v'/2'A B C D E F G H o Om mI Im---- mm mm ol-- r(e

.....,.. .,'.IW ad

FIG. 2. Effect of ,B-globin gene untranslated sequences on vectortransmission. Southern blot analysis of packaging cell (iI-CRE, leftlane) and target cell (NIH 3T3, right lane) genomic DNA. Lanes: A,1.8-kb 13-globin gene and promoter; B, 5' untranslated region deletion;C, intron 1 deletion; D, B and C combined deletions; E, intron 2deletion; F, B and E combined deletions; G, C and E combineddeletions; H, B, C, and E combined deletions; I, 3-globin codingsequence only; J, 770-bp intron 2 deletion. See Fig. 1 legend and textfor sequences and methods.

in lanes J of Fig. 2; and lanes C and F, the globin codingsequence, in the vectors pSG and pMFG, respectively. Com-parison between lanes B and E shows a gain in titer of about5-fold when the pMFG vector is used, showing that alteringvector sequences can affect transmission.The Core Sites of HS2, HS3, and HS4 Can Be Incorporated

into a High-Titer Genomically Stable Retroviral Vector. Pre-vious studies of LCR sequences had indicated that a "minilo-cus" approximately 20 kb in size, comprising sites HS1, HS2,HS3, and HS4, possessed close to complete LCR activity (10,11, 14). However, because of the inability to transmit se-quences of such size in retrovirus vectors, we chose to attemptto incorporate "core elements" of HS2, HS3, and HS4, whichare markedly reduced in size and have been shown to retain atleast partial LCR activity, into vectors carrying the (3-globinstructural gene. The following sequences were employed forour studies, based on studies in transgenic animals (17-22,27-29): (i) the 283-bp Sac I-Ava I core HS4 fragment (29); (ii)the 260-bp core HS3, including the Hph I-Fnu4HI segment(27, 28) and the upstream NF-E2 site, which encompasses allsites footprinted in vivo (45, 46); and (iii) a 478-bp HindIII-SnaBI HS2 fragment encompassing all sites footprinted in vivo(45, 47) in the core HindIII-Xba I fragment (19-22) and mostof the adjacent alternating purine-pyrimidine stretch (21, 48).As shown previously in Fig. 1, one specific orientation of thesites greatly reduced proviral transmission. Accordingly, thethree sites were subcloned in such a way that all eight permu-tations of relative arrangement were generated. The resultingcassettes were introduced, in sense and antisense orientationsupstream of the ,B-globin sequence in the high-titer f3-globinvector shown in lanes B of Fig. 1. As shown in Fig. 4,transmission of these vectors shows significant variations withregard to copy number and genomic stability (e.g., the twospecies transmitted in lanes H). Among these 16 constructs, weidentified four combinations which were transmitted at hightiter without rearrangements (lanes A, E, L, and N). All bearHS2 in the antisense orientation relative to retroviral tran-

1o0 A B C D E F

FIG. 3. Comparison of transmission of 3-globin sequences in twodifferent vectors. Lanes A-C, pSG vector; lanes D-F, pMFG vector.Lanes A and D, 3-globin gene (as in Fig. 1, lanes A); lanes B and E,intron 2 deleted minigene (as in Fig. 2, lanes J); lanes C and F, ,BCS(as in Fig. 1, lanes B). See Figs. 1 and 2 for nomenclature and methods.Signal intensity between lanes can be directly compared owing to veryeven loading (data not shown) except for lanes F, which were loadedwith 5 ,ug.

I J K L M N o ur ---II ~ l 1- -- 0

FIG. 4. Effect of HS orientation on vector transmission. Southernblot analysis of packaging cell (4i-CRE, left lane) and target cell (NIH3T3, right lane) genomic DNA (see previous figure legends formethods). The top band corresponds to the endogenous globin genes.HS orientation (5, sense; s, antisense) in the vector is (in the orderHS4-HS3-HS2): A, Sss; B, reversed A; C, sss; D, reversed C; E, SSs;F, reversed E; G, sSs; H, reversed G; I, SsS; J, reversed I; K, ssS; L,reversed K; M, sSS; and N, reversed M. Lane 3T3, NIH 3T3.

scription. The reason for the lack of the endogenous band inlanes E is unclear.For further studies, we selected the combination with the

highest titer and in which HS2 and HS3 were oriented in thesame direction as the ,B-globin gene and adjacent to itspromoter. The resulting vector, M,B6L, is shown in Fig. 5. Insummary, this construct contains a 2150-bp SnaBI-PstI ,B-glo-bin fragment including a full intron 1 and a 476-bp intron 2 andlacking the 3' enhancer [on the basis of other studies thatsuggest it is redundant in the presence of the LCR (24)]. The1-kb LCR fragment comprises the HS2, HS3, and HS4 coresequences described above (Fig. 5A). Southern blot analysis ofNIH 3T3 cells transduced by M/36L indicates that approxi-mately one proviral copy is transferred per cell (Fig. 5B). Onthe basis of previous comparisons of proviral copy number andtiters measured by selectable marker expression (unpublishedresults), the titer of M136L is approximately 1 X 106 per ml.Mj86L Directs Elevated j8-Globin Expression, Detectable at

All Integration Sites in MEL Cells. To test the biologicalactivity of M136L, we chose MEL cells as target cells for genetransfer, as those cells have been extensively employed instudies of /3-globin transcription (10, 14-16, 23-26). A panel ofMEL cell clones infected with M,B6L was generated by cocul-tivation of C88 MEL cells and virus-producing cells andsubsequent subcloning by limiting dilution (see Materials andMethods). Southern blot analyses confirmed that intact vectorcopies were integrated at different chromosomal positions foreach clone (data not shown). Expression of human ,B-globinwas quantitated by RNase protection using total RNA ex-tracted from dimethyl sulfoxide-induced clones. Levels ofhuman /3-globinmRNA transcripts were measured by using thephosphorimager and normalized to endogenous a- or (3-majorglobin transcripts. This analysis is summarized in Table 1,where values are normalized to human B3-globin expression inHull, a MEL cell containing human chromosome 11 underselective pressure (49), based on the comparison of theirrespective human , to murine a mRNA ratios (calculationsbased on human f3/murine {3-major yielded very similar re-sults). In a pool of 105 unselected MEL cells infected with theparent vector lacking the LCR, M(86, the mean level ofexpression was 2.2% of that measured in Hull. Mean expres-sion was 78% in MEL cell pools infected with M,B6L. Human(-globin transcripts were detected in NIH 3T3 fibroblasts butat levels about 1/100 of the level measured from the equalamount of RNA from induced MEL cells (Table 1). This

6730 Genetics: Sadelain et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 6731

Pst+2792

intron 2deletion SnaB I

- 265

enhancerdeletion

I+ iHS2 HS3 HS4 I B 0.5 1 M16 M,B6L

I IF---l

SD SA690 476 423 280 283 5230

I.*14 IF _jj&.. :.

MFG HUMAN B-GLOBIN GENE LCR SG

FIG. 5. (A) Schematic representation of the M136L vector. Black boxes represent the ,3-globin exons. Arrows above the HS boxes indicate theirorientations. Numbers below delineated segments represent fragment length in bp. 4u+, packaging signal. White boxes at ends represent LTRs. Seetext for exact sequences. SD, splice donor; SA, splice acceptor; MFG, 5' sequences derived from the MFG vector; SG, 3' sequences derived fromthe SG vector. (B) Transmission characteristics of vectors M036L and M136, the parent ,B-globin gene vector lacking the LCR fragment. Southernblot analysis is carried out as in the previous figures. Length on left is in bp.

finding is consistent with other studies of ,B-globin expressionin transfected fibroblasts (14, 15, 23, 24). In subcloned MELcells, the expression values ranged from 4% to 146% (ML1-ML7, Table 1). The level measured for each clone was verystable upon repeated measurement as well as upon reinduc-tion. Importantly, in this set of clones and others with slightlydifferent ,B-globin gene boundaries (MS and RCM, unpub-lished observations), which represent a total of 17 clones,human ,B-globin expression, albeit variable in amount, wasreadily detected in all cases. This is in contrast to clonesinfected with the vector lacking LCR sequences (M36), inwhich human f3-globin expression was undetectable in 3 of 4clones (M1-M4; Table 1).

Table 1. Human ,B-globin expression in dimethyl sulfoxide-inducedMEL cells and in NIH 3T3 fibroblasts

Hug3 RNA/MaHu,B gene RNA per vector

Cell LCR copy number copy, %MEL - 0 0Hull + 1 100MEL/M,B6 - 0.5 2.2MEL/Mg6L + 1.6 783T3/M,86 - 1.9 0.23T3/M36L + 1.8 0.6Ml - 1 2M2 - 1 <0.1M3 - 1 <0.1M4 - 1 <0.1MLl + 1 61ML2 + 1 101ML3 + 1 146ML4 + 1 4ML5 + 3 28ML6 + 4 35ML7 + 5 74

Protected RNA transcripts were fractionated on a 6% polyacryl-amide/urea gel and their radioactivities were measured by using thephosphorimager. Results are expressed as human 13-globin mRNA/endogenous murine a-globin mRNA divided by vector copy number,normalized to the values measured in the Hull clone (see text). Valuesin NIH 3T3 cells are expressed as human ,B-globin RNA divided byvector copy number in 1 ,ug of total RNA normalized to the human,3-globin signal in 1 ,ug of total RNA from dimethyl sulfoxide-inducedHull cells. In the first six lines, data from a large polyclonal cellpopulation; in the rest of the table, data from individual MEL cellclones.

DISCUSSIONIn this study, we report the resolution of two major obstaclesto the implementation of retrovirus-based genetic treatment of,B-globin disorders: first, the generation of a high-titer retro-viral vector suitable for obtaining the efficient transduction ofhematopoietic stem cells and, second, the generation ofvectorscapable of expressing high levels of the f-globin gene in anerythroid-specific fashion. Our systematic study of vectortransmission indicated problems associated with incorporationof both the ,B-globin structural gene itself and the LCR coresequences. In the case of the f3-globin structural gene, ourresults parallel those of Miller et al. (43), who had observedthat the poor transmission of retroviral vectors carrying theP-globin structural gene was caused by both untranslatedsequences in the 5' and 3' regions of the Hpa I-Xba I P3-globinfragment and sequences within intron 2. Our results differslightly from those of that study in that we found that se-quences within intron 2 alone primarily account for the poortransmission of f3-globin vectors (Fig. 2). We found thatfeatures of retroviral vector design also affected proviraltransmission of ,3-globin sequences. For example, modifica-tions of the vector sequences juxtaposed to the 3' end of the13-globin gene indicated that including the Moloney MLVsplice acceptor region from pMFG (34) resulted in a 5-foldincrease in titer. Deletion of the U3 region of the 3' LTR (33)did not decrease titer and, in fact, led to a moderate (2-fold)increase (data not shown).We also confirmed the studies of Novak et al. (30) that

suggested that LCR sequences can lead to significant vectorinstability (data not shown). In our studies, even the core

fragments incorporated together led to decreased transmissionand vector instability in particular sequence arrangements(Figs. 1 and 4). To overcome this problem, we examined thetransmission of all 16 permutations of the core sites, and weidentified combinations that result in stable transmission andthe highest titers (Fig. 4, lanes A, E, L, and N).The LCR was first functionally defined in vivo in transgenic

mice by using a 20-kb LCR fragment (10, 11), a 6.5-kbfragment (15), or single HS sites (17-22, 27-29). Studies inanimals suggested that the LCR confers position-independentexpression, based on the calculation that ,B-globin expressionper transgene copy is relatively constant (7). However, itshould be noted that transgenic studies involving either com-plete LCR or core sequences generally relied on the analysisof cells possessing more than one copy per cell. In contrast, ouruse of retrovirus-mediated gene transfer made it possible toexamine the issue of position independence under conditions

A

- - -- 9- IB 40 0 ----|

Genetics: Sadelain et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995)

where one copy of the transcription unit per cell is integratedin a precise reproducible way and in the absence of anyselection (50). In a panel of MEL cell clones representingindependent integration sites, we found that expression wasdetectable in all clones, yet ranged from 4% to 146% (Table1) of human ,B-globin expression in the Hull cell line insingle-copy clones (the variability decreases, as expected, inmulticopy clones). Our data therefore suggest that the LCR, inthis configuration, acts more like a classical enhancer than theproposed function of an LCR (reviewed in refs. 7-10 and 51).The absence of strictly position-independent expression (10,14, 15) we observe raises the issues of whether the LCR cantruly confer position-independent expression when present atone copy per cell, whether the compact arrangement of thetranscription and chromatin regulators within our vector leadsto suboptimal interactions and inefficient LCR activity, orwhether it is specifically the combination of core elementswhich we have employed which lack this ability (see refs. 17-22for HS2, 27 and 28 for HS3, and 29 for HS4). Whatever theexplanation, our data strongly suggest that it is important tocarefully reevaluate the characteristics of larger LCR-containing sequences in the context of single-copy insertions.For the purposes of gene therapy the most critical test of the

LCR sequences will be the introduction of the virus genomesdescribed above into hematopoietic cells followed by theirtransplantation. The inability to obtain chromosome position-independent expression of the human ,B-globin gene in thecontext of single-copy insertions would make the prospects foreffective genetic treatment of hemoglobinopathies less likelythan previously believed.

M.S. thanks Drs. I. Riviere and M. Goodell for reviewing themanuscript and the Medical Research Council of Canada for support.M.S. is a recipient of the Medical Research Council CentennialFellowship. C.H.J.W. was supported by the Undergraduate ResearchOpportunities Program at the Massachusetts Institute of Technology.This work was supported by National Institutes of Health GrantHL37569 to R.C.M.

1. Dzierzak, E. A., Papayannopoulou, T. & Mulligan, R. C. (1988)Nature (London) 331, 35-41.

2. Karlsson, S., Bodine, D. M., Perry, L., Papayannopoulou, T. &Nienhuis, A. W. (1988) Proc. Natl. Acad. Sci. USA 85, 6062-6066.

3. Bodine, D. M., Karlsson, S. & Nienhuis, A. W. (1989) Proc. Natl.Acad. Sci. USA 86, 8897-8901.

4. Bender, M. A., Gelinas, R. E. & Miller, A. D. (1989) Mo. Cell.Bio. 9, 1426-1434.

5. Tuan, D., Solomon, W., Li, Q. & London, I. M. (1985) Proc. Natl.Acad. Sci. USA 82, 6384-6388.

6. Forrester, W. C., Takegawa, S., Papayannopoulou, T., Stamatoy-annopoulos, G. & Groudine, M. (1987) Nucleic Acids Res. 15,10159-10177.

7. Dillon, N. & Grosveld, F. (1993) Trends Genet. 9, 134-137.8. Crossley, M. & Orkin, S. H. (1993) Curr. Opin. Genet. Dev. 3,

232-237.9. Engel, J. D. (1993) Trends Genet. 9, 304-309.

10. Grosveld, F., van Assendelft, G. B., Greaves, D. R. & Kollias, G.(1987) Cell 51, 975-985.

11. Ryan, T. M., Behringer, R. R., Martin, N. C., Townes, T. M.,Palmiter, R. D. & Brinster, R. L. (1989) Genes Dev. 3, 314-323.

12. Peterson, K. R., Clegg, C. H., Huxley, C., Josephson, B. M.,Haugen, H. S., Furukawa, T. & Stamatoyannopoulos, G. (1993)Proc. Natl. Acad. Sci. USA 90, 7593-7597.

13. Gaensler, K. M. L., Kitamura, M. & Kan, Y. W. (1993) Proc.Natl. Acad. Sci. USA 90, 11381-11385.

14. van Assendelft, G. B., Hanscombe, O., Grosveld, F. & Greaves,D. R. (1989) Cell 56, 969-977.

15. Talbot, D., Collis, P., Antoniou, M., Vidal, M., Grosveld, F. &Greaves, D. R. (1989) Nature (London) 338, 352-355.

16. Fraser, P., Hurst, J., Collis, P. & Grosveld, F. (1990) NucleicAcidsRes. 18, 3503-3508.

17. Ryan, T. M., Behringer, R. R., Townes, T. M., Palmiter, R. D. &Brinster, R. L. (1989) Proc. Natl. Acad. Sci. USA 86, 37-41.

18. Curtin, P. T., Liu, D., Liu, W., Chang, J. C. & Kan, Y. W. (1989)Proc. Natl. Acad. Sci. USA 86, 7082-7086.

19. Talbot, D., Philipsen, S., Fraser, P. & Grosveld, F. (1990) EMBOJ. 9, 2169-2178.

20. Caterina, J. J., Ryan, T. M., Pawlik, K. M., Palmiter, R. D.,Brinster, R. L., Behringer, R. R. & Townes, T. M. (1991) Proc.Natl. Acad. Sci. USA 88, 1626-1630.

21. Liu, D. P., Chang, J. C., Moi, P., Liu, W., Kan, Y. W. & Curtin,P. T. (1992) Proc. Natl. Acad. Sci. USA 89, 3899-3903.

22. Caterina, J. J., Ciavatta, D. J., Donze, D., Behringer, R. R. &Townes, T. M. (1994) Nucleic Acids Res. 22, 1006-1011.

23. Forrester, W. C., Novak, U., Gelinas, R. & Groudine, M. (1989)Proc. Natl. Acad. Sci. USA 86, 5439-5443.

24. Collis, P., Antoniou, M. & Grosveld, F. (1990) EMBO J. 9,233-240.

25. Ney, P. A., Sorrentino, B. P., McDonagh, K. T. & Nienhuis,A. W. (1990) Genes Dev. 4, 993-1006.

26. Talbot, D. & Grosveld, F. (1991) EMBO J. 10, 1391-1398.27. Philipsen, S., Talbot, D., Fraser, P. & Grosveld, F. (1990)EMBO

J. 9, 2159-2167.28. Philipsen, S., Pruzina, S. & Grosveld, F. (1993) EMBO J. 12,

1077-1085.29. Pruzina, S., Hanscombe, O., Whyatt, B., Grosveld, F. & Phil-

ipsen, S. (1991) Nucleic Acids Res. 19, 1413-1419.30. Novak, U., Harris, E. A. S., Forrester, W., Groudine, M. &

Gelinas, R. (1990) Proc. Natl. Acad. Sci. USA 87, 3386-3390.31. Plavec, I., Papayannopoulou, T., Maury, C. & Meyer, F. (1993)

Blood 81, 1384-1392.32. Chang, J. C., Liu, D. & Kan, Y. W. (1992) Proc. Natl. Acad. Sci.

USA 89, 3107-3110.33. Guild, B. C., Finer, M. H., Housman, D. H. & Mulligan, R. C.

(1988) J. Virol. 62, 3795-3801.34. Dranoff, G., Jaffee, E., Lazenby, A., Golumbek, P., Levitsky, H.,

Brose, K., Jackson, V., Hamada, H., Pardoll, D. & Mulligan,R. C. (1993) Proc. Natl. Acad. Sci. USA 90, 3539-3543.

35. Southern, P. J. & Berg, P. (1982) J. Mol. Appl. Genet. 1, 327-341.36. Danos, 0. & Mulligan, R. C. (1988) Proc. Natl. Acad. Sci. USA

85, 6460-6464.37. Auffray, C. & Rougeon, F. (1980) Eur. J. Biochem. 107, 303-314.38. Zinn, K., DiMaio, D. & Maniatis, T. (1983) Cell 34, 865-879.39. Charnay, P., Treisman, R., Mellon, P., Chao, M., Axel, R. &

Maniatis, T. (1984) Cell 38, 251-263.40. Nishioka, Y. & Leder, P. (1979) Cell 18, 875-882.41. Buchman, A. R. & Berg, P. (1988) Mol. Cell. Biol. 8, 4395-4405.42. Bender, M. A., Miller, A. D. & Gelinas, R. E. (1988) Mol. Cell.

Biol. 8, 1725-1735.43. Miller, A. D., Bender, M. A., Harris, E. A. S., Kaleko, M. &

Gelinas, R. E. (1988) J. Virol. 62, 4337-4345.44. Liu, X. & Mertz, J. E. (1993) Nucleic Acids Res. 21, 5256-5263.45. Ikuta, T. & Kan, Y. W. (1991) Proc. Natl. Acad. Sci. USA 88,

10188-10192.46. Strauss, E. C. & Orkin, S. H. (1992) Proc. Natl. Acad. Sci. USA

89, 5809-5813.47. Reddy, P. M. S. & Shen, C.-K. J. (1991) Proc. Natl. Acad. Sci.

USA 88, 8676-8680.48. Moi, P. & Kan, Y. W. (1990) Proc. Natl. Acad. Sci. USA 87,

9000-9004.49. Nandi, A. K., Roginski, R. S., Gregg, R. G., Smithies, 0. &

Skoultchi, A. I. (1988) Proc. Natl. Acad. Sci. USA 85, 3845-3849.50. Miller, J. L., Walsh, C. E., Ney, P. A., Samulski, R. J. & Nienhuis,

A. W. (1993) Blood 82, 1900-1906.51. Felsenfeld, G. (1992) Nature (London) 355, 219-224.

6732 Genetics: Sadelain et aL

Dow

nloa

ded

by g

uest

on

Janu

ary

16, 2

020